mathlib3 documentation

topology.metric_space.gromov_hausdorff_realized

The Gromov-Hausdorff distance is realized #

THIS FILE IS SYNCHRONIZED WITH MATHLIB4. Any changes to this file require a corresponding PR to mathlib4.

In this file, we construct of a good coupling between nonempty compact metric spaces, minimizing their Hausdorff distance. This construction is instrumental to study the Gromov-Hausdorff distance between nonempty compact metric spaces.

Given two nonempty compact metric spaces X and Y, we define optimal_GH_coupling X Y as a compact metric space, together with two isometric embeddings optimal_GH_injl and optimal_GH_injr respectively of X and Y into optimal_GH_coupling X Y. The main property of the optimal coupling is that the Hausdorff distance between X and Y in optimal_GH_coupling X Y is smaller than the corresponding distance in any other coupling. We do not prove completely this fact in this file, but we show a good enough approximation of this fact in Hausdorff_dist_optimal_le_HD, that will suffice to obtain the full statement once the Gromov-Hausdorff distance is properly defined, in Hausdorff_dist_optimal.

The key point in the construction is that the set of possible distances coming from isometric embeddings of X and Y in metric spaces is a set of equicontinuous functions. By Arzela-Ascoli, it is compact, and one can find such a distance which is minimal. This distance defines a premetric space structure on X ⊕ Y. The corresponding metric quotient is optimal_GH_coupling X Y.

def Gromov_Hausdorff.candidates (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

set (prod_space_fun X Y)

The set of functions on X ⊕ Y that are candidates distances to realize the minimum of the Hausdorff distances between X and Y in a coupling

Equations

Gromov_Hausdorff.candidates X Y = {f : prod_space_fun X Y | (((((∀ (x y : X), f (sum.inl x, sum.inl y) = has_dist.dist x y) ∧ ∀ (x y : Y), f (sum.inr x, sum.inr y) = has_dist.dist x y) ∧ ∀ (x y : X ⊕ Y), f (x, y) = f (y, x)) ∧ ∀ (x y z : X ⊕ Y), f (x, z) ≤ f (x, y) + f (y, z)) ∧ ∀ (x : X ⊕ Y), f (x, x) = 0) ∧ ∀ (x y : X ⊕ Y), f (x, y) ≤ ↑(max_var X Y)}

def Gromov_Hausdorff.candidates_b_of_candidates {X : Type u} {Y : Type v} [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] (f : prod_space_fun X Y) (fA : f ∈ Gromov_Hausdorff.candidates X Y) :

Cb X Y

candidates give rise to elements of bounded_continuous_functions

Equations

Gromov_Hausdorff.candidates_b_of_candidates f fA = bounded_continuous_function.mk_of_compact {to_fun := f, continuous_to_fun := _}

theorem Gromov_Hausdorff.candidates_b_of_candidates_mem {X : Type u} {Y : Type v} [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] (f : prod_space_fun X Y) (fA : f ∈ Gromov_Hausdorff.candidates X Y) :

Gromov_Hausdorff.candidates_b_of_candidates f fA ∈ candidates_b X Y

noncomputable def Gromov_Hausdorff.candidates_b_dist (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [inhabited X] [metric_space Y] [compact_space Y] [inhabited Y] :

Cb X Y

The distance on X ⊕ Y as a candidate

Equations

Gromov_Hausdorff.candidates_b_dist X Y = Gromov_Hausdorff.candidates_b_of_candidates (λ (p : (X ⊕ Y) × (X ⊕ Y)), has_dist.dist p.fst p.snd) _

theorem Gromov_Hausdorff.candidates_b_dist_mem_candidates_b {X : Type u} {Y : Type v} [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

Gromov_Hausdorff.candidates_b_dist X Y ∈ candidates_b X Y

noncomputable def Gromov_Hausdorff.HD {X : Type u} {Y : Type v} [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] (f : Cb X Y) :

We will then choose the candidate minimizing the Hausdorff distance. Except that we are not in a metric space setting, so we need to define our custom version of Hausdorff distance, called HD, and prove its basic properties.

Equations

Gromov_Hausdorff.HD f = linear_order.max (⨆ (x : X), ⨅ (y : Y), ⇑f (sum.inl x, sum.inr y)) (⨆ (y : Y), ⨅ (x : X), ⇑f (sum.inl x, sum.inr y))

theorem Gromov_Hausdorff.HD_below_aux1 {X : Type u} {Y : Type v} [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] {f : Cb X Y} (C : ℝ) {x : X} :

bdd_below (set.range (λ (y : Y), ⇑f (sum.inl x, sum.inr y) + C))

theorem Gromov_Hausdorff.HD_below_aux2 {X : Type u} {Y : Type v} [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] {f : Cb X Y} (C : ℝ) {y : Y} :

bdd_below (set.range (λ (x : X), ⇑f (sum.inl x, sum.inr y) + C))

theorem Gromov_Hausdorff.HD_candidates_b_dist_le {X : Type u} {Y : Type v} [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

Gromov_Hausdorff.HD (Gromov_Hausdorff.candidates_b_dist X Y) ≤ metric.diam set.univ + 1 + metric.diam set.univ

Explicit bound on HD (dist). This means that when looking for minimizers it will be sufficient to look for functions with HD(f) bounded by this bound.

noncomputable def Gromov_Hausdorff.premetric_optimal_GH_dist (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

pseudo_metric_space (X ⊕ Y)

With the optimal candidate, construct a premetric space structure on X ⊕ Y, on which the predistance is given by the candidate. Then, we will identify points at 0 predistance to obtain a genuine metric space

Equations

Gromov_Hausdorff.premetric_optimal_GH_dist X Y = {to_has_dist := {dist := λ (p q : X ⊕ Y), ⇑(optimal_GH_dist X Y) (p, q)}, dist_self := _, dist_comm := _, dist_triangle := _, edist := λ (x y : X ⊕ Y), ↑⟨(λ (p q : X ⊕ Y), ⇑(optimal_GH_dist X Y) (p, q)) x y, _⟩, edist_dist := _, to_uniform_space := uniform_space_of_dist (λ (p q : X ⊕ Y), ⇑(optimal_GH_dist X Y) (p, q)) _ _ _, uniformity_dist := _, to_bornology := bornology.of_dist (λ (p q : X ⊕ Y), ⇑(optimal_GH_dist X Y) (p, q)) _ _ _, cobounded_sets := _}

@[nolint]

def Gromov_Hausdorff.optimal_GH_coupling (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

Type (max u v)

A metric space which realizes the optimal coupling between X and Y

Equations

Gromov_Hausdorff.optimal_GH_coupling X Y = uniform_space.separation_quotient (X ⊕ Y)

Instances for Gromov_Hausdorff.optimal_GH_coupling

@[protected, instance]

noncomputable def Gromov_Hausdorff.optimal_GH_coupling.metric_space (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

metric_space (Gromov_Hausdorff.optimal_GH_coupling X Y)

noncomputable def Gromov_Hausdorff.optimal_GH_injl (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] (x : X) :

Gromov_Hausdorff.optimal_GH_coupling X Y

Injection of X in the optimal coupling between X and Y

Equations

Gromov_Hausdorff.optimal_GH_injl X Y x = quotient.mk' (sum.inl x)

theorem Gromov_Hausdorff.isometry_optimal_GH_injl (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

isometry (Gromov_Hausdorff.optimal_GH_injl X Y)

The injection of X in the optimal coupling between X and Y is an isometry.

noncomputable def Gromov_Hausdorff.optimal_GH_injr (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] (y : Y) :

Gromov_Hausdorff.optimal_GH_coupling X Y

Injection of Y in the optimal coupling between X and Y

Equations

Gromov_Hausdorff.optimal_GH_injr X Y y = quotient.mk' (sum.inr y)

theorem Gromov_Hausdorff.isometry_optimal_GH_injr (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

isometry (Gromov_Hausdorff.optimal_GH_injr X Y)

The injection of Y in the optimal coupling between X and Y is an isometry.

@[protected, instance]

def Gromov_Hausdorff.compact_space_optimal_GH_coupling (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] :

compact_space (Gromov_Hausdorff.optimal_GH_coupling X Y)

The optimal coupling between two compact spaces X and Y is still a compact space

theorem Gromov_Hausdorff.Hausdorff_dist_optimal_le_HD (X : Type u) (Y : Type v) [metric_space X] [compact_space X] [nonempty X] [metric_space Y] [compact_space Y] [nonempty Y] {f : Cb X Y} (h : f ∈ candidates_b X Y) :

metric.Hausdorff_dist (set.range (Gromov_Hausdorff.optimal_GH_injl X Y)) (set.range (Gromov_Hausdorff.optimal_GH_injr X Y)) ≤ Gromov_Hausdorff.HD f

For any candidate f, HD(f) is larger than or equal to the Hausdorff distance in the optimal coupling. This follows from the fact that HD of the optimal candidate is exactly the Hausdorff distance in the optimal coupling, although we only prove here the inequality we need.